Contact: Isabelle Guyon - Submitted: 2008-09-12 02:53 - Views : 1651 - [Edit entry]

• Authors: Causality workbench team
• Key facts: Number of variables: 4932 (molecular descriptors) + one binary target variable (molecular activity).
  Number of examples: training 12678 + 3 test sets of 10000 examples corresponding to unmanipulated data and 2 different types of manipulations.
  Variable types: binary. Some variables are real and other artificial (probes).
  Missing data: no.
• Keywords: probe.method, pharmacology, QSAR
• Download BibTeX
• Download the dataset
• Pot luck task proposed: LOCANET

Abstract:

This is one of the datasets of the first causality challenge: causation and prediction. The goal of the challenge was to make predictions under manipulations.

SIDO (SImple Drug Operation mechanisms) contains descriptors of molecules, which have been tested against the AIDS HIV virus. The target values indicate the molecular activity (+1 active, -1 inactive). The causal discovery task is to uncover causes of molecular activity among the molecule descriptors. This would help chemists in the design of new compounds, retaining activity, but having perhaps other desirable properties (less toxic, easier to administer). This dataset is semi-artificial: it contains both real variables and artificial variables (probes).

For the pot-luck challenge, the task proposed is to discover the causal network in the neighborhood of the target.

#1

Montassar Ben messaoud

2008-10-15 12:40:48

How can i find the cost of each manipulation please?